International Journal of Computer Sciences and Engineering Vol.

9(4), Apr 2021, E-ISSN: 2347-2693

Employee recruitment and other counseling sessions will

be done based on personality assessment. A person should
know about their personality by conducting different
personality tests [6].

A Personality test is like observation of psychologists or

self-descriptive report. These are costly traditional methods
and less practical. Recently, the personality test is
conducting using a website like online questionnaire and
this is quite practical. In this type of test the user has to
answer various questions. A recent study saying that from
the text written by the user, we can obtain the personality
traits [7]. The choice of most frequently used words can
describe the personality of that particular person. Social
media is a platform where represent users to the world by
them. Personality classifies a human that affects
preferences of an individual and interaction. To find out
their personality, people are required to take a personality
test. Social media became a platform for people to express
Figure 1: Personality Traits
their feelings to the world, based on the posts in social
media made by users can be used to analyze their behavior
From the previous analysis 14 out of 60 items can be
[8]. This experiment uses text written by Twitter users to
replaced in the NEO Five Factor Inventory. For the new
predict personality of the particular user. The used are
analysis in high school (N=1960) and adult (N=1493)
English and Indonesian. Relations can be presented in
samples have been taken from the remaining set of revised
among user preferences in different entertainment domains
NEO Personality Inventory items. The results show the
and personality types. The entertainment domains namely
reliability of personality traits. These accurate results are
movies, music, TV shows and books. For analyzing we
appropriate for the age 16 and up in responders [11].
have taken a total of 53,224 Twitter user profiles. These
Although, continued use of the present tool is also fair and
profiles were composed of both personality scores from the
sensible for many applications. The NEO Five Factor
Five Factor model (openness, conscientiousness, Inventory was not considered to supply definitive amount
extraversion, agreeableness, neuroticism) and explicit
of the 5 personality factors. Alternatively, it was designed
interests in each of the above domains [9]. From our
for investigative research as a brief tool which is
analysis, we extract different personality stereotypes,
reasonable for personality factors [12]. This is used in most
similarity index of the personality with the personality of
applications. Rest of the article is organized as follows:
different domain and association rules for some of the
Related works are discussed in section 2, the proposed
domains.
methodology is explained in section 3, section 4 presented
a experimental results and finally section 5 presents a
Personality is most important and complex attribute among conclusion.
other four attributes like behavioral, emotional,
temperamental and mental [10]. This personality describes
II. RELATED WORK
a unique individual. We know that words used by a person
define the information about the speaker and their semantic In author proposed another approach for personality
content. This type of data contains cues of speaker‟s
discovery which depends on joining the opinion, full of
personality traits. The Big Five can be used to assess the
feeling and presence of mind learning from the utilizing
personality and shown in Figure 1.
assets viz. Sentic-Net, Concept-Net, Emo-Sentic-Net and
Emo-Sentic-Space [13]. In their approach, they
• Introversion vs. Extraversion (reserved, aloof, shy vs. consolidated good judgment information based highlights
playful, sociable, assertive) with psycho-phonetic highlights and recurrence based
• Disagreeable vs. Agreeableness (faultfinding vs.
highlights and later the highlights were utilized in directed
cooperative)
classifiers. Further, they created five help vector machine
• Neuroticism vs. Emotional stability (insecure vs.
models for five personality attributes. Their exploratory
unemotional) outcomes demonstrate that the utilization of presence of
• Unconscientiously vs. Conscientiousness (careless vs. mind learning with emotional and feeling information
self-disciplined)
upgrades the exactness of the current systems which utilize
• Openness to experience (unimaginative)
just psycho-phonetic highlights and recurrence based
investigation at lexical level [14]. In writers display a
programmed personality attribute acknowledgment show
in light of social system (Face book) utilizing users' status.
They utilized machine learning calculations viz. Bolster

© 2021, IJCSE All Rights Reserved 8

 International Journal of Computer Sciences and Engineering
Vector Machine, Bayesian Logistic Regression (BLR) and
Multinomial Naïve Bayes (MNB). In author created three
machine learning calculations i.e. bolster vector machine,
nearest neighbor with k=1 (kNN) and Naïve Bayes for
deducing the personality attributes of users on the premise
of their face book refreshes. A few characterization
procedures were utilized to manufacture prescient
personality models along the five personality
measurements utilizing the etymological highlights of a
dataset involved couple of thousand s requested from basic
brain science understudies. Authors in fabricated
personality acknowledgment show in both discussion and
by means of Big. They abused two lexical assets as
highlights, LIWC and MRC, and anticipated both
personality scores and classes utilizing Support Vector
Machines (SVMs) and M5 trees individually. They
likewise detailed an extensive rundown of connections
between Big5 personality characteristics and two lexical
assets they utilized [15].
The Linguistic Inquiry and Word Count – LIWC
(http://www.liwc.net) was utilized as an apparatus for
etymological examination. In author took after the work
exhibited indirect and built up personality recognition
demonstrate for present day Greek with etymological
highlights (like Part-of-Speech labels) and mental
highlights (like in LIWC). They utilized SVM classifier for
building the machine learning model, they exhibited that
personality and dialect can be effectively ported from
English to different dialects. In author utilized n-gram
highlights from width of online journals a corpus of
individual web-s for displaying four out of five personality
measurements [16]. They manufactured their model with
SMO and Naïve Bayes machine learning techniques. Their
outcomes call attention to the significance of the
component choice in expanding the classifiers exactness
yielding 83%-93% for programmed includes choice. The
relationship between users' personality and social system
action has been the concentration of a few investigations in
the last. In author removed word n-grams as highlights
from a vast corpus of online journals with various element
vector development settings, for example, the
nearness/nonappearance of stop words or converse record
recurrence. They found that bigrams, regarded as Boolean
highlights and keeping stop words, yield great outcomes
utilizing SVMs as learning calculation. Golbecketal
proposed a model to anticipate personality from Facebook
profile with semantic, (for example, word tally) and social
system highlights (like companions check) information
utilizing machine learning calculations [17]. They
anticipated personality scores of 279 Facebook users,
misusing both semantic highlights (from LIWC) and social
highlights (i.e. companion tally, relationship status). In
author additionally anticipated the personality of 279
Twitter users with the assistance of LIWC, auxiliary
highlights (i.e. hastags, connections) and assumption
highlights, and utilizing a Gaussian Process (GP) as
learning calculation. Tomlinson et al. contemplated the
Conscientiousness quality to recognize objective,
© 2021, IJCSE All Rights Reserved
Vol.9(4), Apr 2021, E-ISSN: 2347-2693
inspiration, and the way the creator sees control over the
depicted circumstances. They played out the examination
of occasion structures of printed client announcements in a
Facebook dataset In writers utilized system highlights
(like adherents, following, and so on.) to assemble M5
rules based learning model for the expectation of
personality scores of 335 Twitter users [18].
Authors in introduced a broad investigation of the system
qualities (i.e. for example, size of companionship organize,
transferred photographs, occasions went to, times client
has been labeled in photographs) that relate with
personality of 180000 Facebook users. They anticipated
personality scores utilizing multivariate direct relapse
(mLR), and detailed great outcomes on extraversion [19].
Ross et al. spearheaded the investigation of the connection
amongst personality and examples of social system utilize.
They guessed numerous connections amongst personality
and Facebook highlights, including (1) constructive
connection amongst Extraversion and Facebook utilize,
number of Facebook companions and relationship with
Facebook gatherings; (2) constructive connection amongst
Neuroticism and uncovering private information on
Facebook; (3) constructive relationship amongst
Agreeableness and number of Facebook companions; (4)
constructive connection amongst Openness and number of
various Facebook highlights utilized; (5) adverse
connection amongst Conscientiousness and general
utilization of Facebook. In author built up a machine
learning model for neuroticism and extraversion utilizing
semantic highlights, for example, work words, deictic,
evaluation articulations and modular verbs. In author
utilized different feeling dictionaries like NRC hash label
feeling vocabulary and NRC feeling dictionary for the
personality identification and discovered key change in the
exactness of the PRT framework [20].
III. PROPOSED METHODOLOGY
The proposed system can gather set of tweets from
persons. After that this text can be processed into vector
data. Stratification method will categorize user‟s text into a
considered data set. The test results were predictions for
each and every Big Five Personality traits [21]. The
primary and secondary personality characteristics were
gained from the amalgamation of two traits. The system
proposed was a web application.
3.1 Personality Detection Model
Information mining procedures assume an essential part in
removing connection designs amongst personality and
assortment of user‟s information caught from different
sources. For the most part, two methodologies were
embraced for examining personality qualities of social
system users. The primary approach utilizes an assortment
of machine learning in view of social calculations to
fabricate models system activities just. The second one
expands the personality-related highlights with
etymological prompts [22].
9
 International Journal of Computer Sciences and Engineering
This intrigue is because of the way personality recognition
that is likewise extremely valuable in social system
investigation and supposition mining that is huge and
creating fields of research. Online social systems are
composed of immense archives information which is
appropriate for personality acknowledgment; still, there are
a few issues in utilizing them for building such models. (1)
Social system information is for the most part not freely
accessible, (2) if information is unlabeled, (3) it is
extremely hard to comment on with personality judgments
and (4) Generally, it is in a variety of dialects. Concept
diagram of proposed model is shown in figure 2.
Data Collectios
Preprocessing
(Twetts)
Transformation Classfic
ation
Figure 2: User Vs Personality Trait
1) Data Collection: For exhibiting the framework, we
require tweets posted by an individual(s). For this, tweets
are acquired utilizing Twitter API. Twitter API gives an
entrance to twitter information containing information
about the users, tweets distributed by a client, list items on
twitter and so on. Tweet protest is in .json organized.
2) Pre-handling module: The tweets are first gotten from
the tweet protest. The framework at that point removes
meta-traits from the tweets. The information separated can
be isolated into social behavior and syntactic information.
3) Transformation module: This module changes the
"multi-mark issue into twofold arrangement issues". This
module gets the Meta attributes removed from the past
module. Utilizing this information, it develops an element
vector. Each position in the vector compares to a meta-
quality.
4) Classification module: A Multilayer Perceptron (MLP)
Neural Network is utilized for characterization. There are
five neural systems (classifiers), one for every personality
characteristic. The yield of every classifier is either '1'
(yes) or '0' (no) contingent upon whether the vectors
coordinates or not, further surmising that the individual has
the personality quality or number.
In this segment, we propose our system after significant
investigation of past work done in this field and bits of
knowledge in light of writing review of particular research
papers in setting to personality forecast from Twitter,
which includes a personality expectation show in view of
Logistic Regression Classifier. Since, personality
expectation is an arrangement issue, our model advances
another approach of anticipating one's personality utilizing
Logistic Regression calculation with a limited blunder
© 2021, IJCSE All Rights Reserved
Vol.9(4), Apr 2021, E-ISSN: 2347-2693
work utilizing stochastic inclination drop which tends to
specific impediments that ran over to specialists already
[23]. The usage of our model is depicted in a diagrammatic
flowchart of different advances performed in this
examination. This approach conveys the possibility to
defeat the disadvantages of past work done. We have
additionally assessed our model with various assessment
measurements, for example, exactness, review and
precision and furthermore contrasted our outcomes and
Naïve Bayes Classifier utilized preceding foresee user‟s
personality by mapping phonetic highlights of tweets into
various personality classes to which a client might possibly
follow and architecture of the proposed model shown in
figure 3.
Structural Features
Through a Twitter application, we are able to collect
information about the user‟s egocentric network. We first
obtained a list of friends. We were interested in density,
and Twitter provides some information about links
between a user‟s friends. A separate query must be made
for each pair of users to determine if they are or are not
friends. It was not possible to submit a query for each pair
of friends because the Twitter application would timeout;
Twitter limits the time an application can run, and since
each query is sent over the network, performance becomes
an issue [24]. Thus, we sampled 2,000 unique pairs of
friends from a user‟s egocentric network and used that to
determine the density of the network, i.e. what percentage
of possible edges between friends exist.
Personal Information
Users provide a wealth of personal information. We
collected everything available, even though some features
would turn out to have no use in our analysis. The raw data
included features like the user‟s name, birthday,
relationship status, religion, education history, gender, and
hometown. Most of this information was not required, so
some users did not in clude all information. Where
possible, we created additional features that indicated
whether or not the user had included the information (e.g.
was a religion or hometown provided or not), or how many
items were listed (e.g. how many educational experiences
were listed). These added features turned out to be much
more useful and predictive than the original raw data. For
example, from 279 users, 111 listed a religion. Within
those 111 people were 82 different entries. This creates a
space too sparse to do any statistical analysis, but just
knowing if a person listed a religion or not reveals insights
into what they are willing to share.
10
 International Journal of Computer Sciences and Engineering Vol.9(4), Apr 2021, E-ISSN: 2347-2693

analysis methods to help predict personality. However, the

text samples used in earlier studies are much larger than
are available to us through Twitter. Data collected in was
used in both studies mentioned above. They had three
separate sources of text, ranging from an average of 1,770
words to over 5,000 words per person. We are in a much
different position pulling text from Twitter profiles. We
combined status updates, the ”About Me” text, and the
”blurb” text from the profile into a single string so there
would be enough text to analyze. The average number of
words in these three strings combined was 26.6. Many
users did not have enough words for linguistic analysis.
Fifty-four subjects had no text at all in these three fields,
and 112 had fewer than 10 words. We eliminated these
users to the text analysis statistics would not be too noisy.
This left us with 167 subjects with an average of 42.6
words per person. Note that the elimination of subjects
with too little text should have a limited impact on our
results. For each personality factor, a two-tailed Student‟s
t-test showed no significant difference in the personality
score between users with 10 words or more and users with
fewer than 10. Following the methods used in as well as
other studies of Twitter behavior, such as, we utilized the
Linguistic Inquiry and Word Count (LIWC) tool to analyze
the text. LIWC produces statistics on 81 different features
of text in five categories. These include Standard Counts
(word count, words longer than six letters, number of
prepositions, etc.), Psychological Processes (emotional,
cognitive, sensory, and social processes), Relativity (words
about time, the past, the future), Personal Concerns (such
as occupation, financial issues, health), and Other
dimensions [26].
Figure 3: Proposed Architecture Diagram
Internal Twitter Statistics
Activities and Preferences
A number of features were available that described a user‟s
Providing lists of personal activities or favorite things has
experience, settings, and history with Twitter. This
always been a part of Twitter. Users list favorite TV
included the user ID, an integer value that corresponds to
shows, movies, music, book, quotes, as well as political
when the user joined the network (lower values indicate an
and organizational affiliations and favorite activities. As
earlier join time), the unix timestamp of their last profile
was the case with religion described above, the space is far
update, the number of notes (short messages) posted, and
to sparse to do any analysis over the actual entries in these
other features that proved less useful such as the URL of
fields, so we created more companion measures. For lists
the profile picture, whether or not their profile was blocked
of favorite things and activities, we counted the number of
(no one‟s was), whether the person was an app user
characters in the entry, roughly measuring how much
(everyone was), and if they had provided a status update
information the user provided in each field. This included
(everyone had). Personality and Profile Correlations We
values of 0 for users who did not supply any information.
had 279 subjects who completed the personality inventory,
For organizational affiliations, we counted the number
but we only used data from the 167 subjects who had at
listed and for political affiliations, we simply measured
least 10 words among all of their text fields so we could
whether it was shared or not [25].
perform a linguistic analysis. Demographic information
was pulled from their Twitter profiles. Among these
Language Features
subjects, the average age was 31.2 years (std dev 8.7). Of
Similar to the activities and likes described above, users
those reporting gender, 68 were female and 61 were male
also have opportunities to share more personal written
(38 did not report). In terms of location, 82.6% (138) were
information through the ”About Me” and ”blurb” text in
from the United States with the remaining subjects coming
their profiles, and through status updates. We collected
from India (8), Australia (7), Italy (7), and others (7). We
these entries and also added features to measure the
found many weak correlations between users‟ profile
character length of each entry. Previous research has
features and personality scores. This echos previous results
shown that linguistic features can be used to predict
of linguistic analysis and personality found in [27]. These
personality traits. Since there is text available on users‟
are reported in table and statistically significant
Twitter profiles, there is potential to apply these linguistic

© 2021, IJCSE All Rights Reserved 11

 International Journal of Computer Sciences and Engineering Vol.9(4), Apr 2021, E-ISSN: 2347-2693

correlations (p < 0.05) are bolded. Below, we discuss some techniques that benefit from this knowledge of consumer
of the more interesting relationships. background. Recommender systems may also benefit from
integrating predicted personality values. Results showing
Predicting Personality correlations between personality and music taste are well
Our feature set for each user included all meaningful established in the literature. Inferring personality traits
features. We excluded those which could not be quantified from Twitter profiles may allow recommender systems to
(e.g. picture URL), for which the value was the same for improve their accuracy by recommending music, and
all users (e.g. if their profile was blocked), or where the possibly other items, that are tailored to the user‟s
data was so sparse that it would not be predictive (e.g. personality profile.
personal website URL). Where possible, we included our
companion statistics on these features (e.g. while the actual IV. RESULTS AND DISCUSSION
website URL was not used, a feature indicating presence or
absence of the URL was included). Linguistic features By and by, to specify monotonously, that personality
were included as described above. We also added five expectation from tweets still goes under semantic
additional features. We ran a multiple linear regression examination and is available to even now an immense
analysis for each personality factor, producing a vector of range of potential outcomes to enhance its course of nature
weights for each feature. The dot product of the weight after some time as more institutionalization will be
vector and the feature vector was computed for each user presented in removing significant surmising from content.
and for each personality feature to create five composite This approach supplements explore business related to not
features. In total, we had 74 features per user. To predict just versatile expectations utilizing machine learning and
the score of a given personality feature, we performed a Twitter yet additionally proposal methodologies for users
regression analysis in Weka with a 10-fold cross- and customized client involvement with promotions and
validation with 10 iterations using two algorithms: substance recommendation. However, our working usage
M5′Rules, a rule-based variation of the M5′ algorithm [28], of the personality expectation demonstrates utilizing
and Gaussian Processes. NLTK toolbox, Scikit-learn and Python can be informed
with the accompanying forbidden information.
Discussions Comparative Analysis of proposed model with Naïve
The question that arises from this research is how the Bayes is displayed in Table 1 and graphs are shown in
results can be used. Drawing on research results that figure 4 precision, accuracy and recall respectively.
connect personality type to behavior and preferences, there
is potential to integrate previous personality results into Table 1: Comparative analysis of Naïve Bayes and proposed
social media as a way to enhance the accuracy of certain model
features or the user‟s experience. Research on interface Naïve Bayes Proposed Model
preference and personality type showed that users No. of Model (Logistic Regression)
preferred interfaces designed to represent personalities that features
P A R P A R
most closely matched their own. This has significant
implications for this work. With the ability to infer a user‟s 100 64.91 68.91 66.91 65.21 70.25 67.26
personality, social media websites, e-commerce retailers,
and even ad servers can be tailored to reflect the user‟s 500 66.48 71.48 68.28 67.21 72.24 68.43
personality traits and present information such that users
will be most receptive to it. For example, the presentation 1000 68.39 74.39 69.79 69.45 76.78 69.99
of Twitter ads could be adjusted based on the personality
of the user. Similarly, product reviews from authors with
personality traits similar to the user could be highlighted to 1500 71.47 79.21 72.47 72.21 81.21 73.19
increase trust and perceived usefulness by the user.
Customized website “skins” could be created for different
10000 76.36 83.36 78.36 77.23 84.45 78.97
user personality types, as suggested.

Our methods provide a straightforward way to obtain 15000 78.89 85.19 79.99 81.92 87.84 82.64
personality profiles of users without the burden of tests,
and this will make it much easier to create personality-
20000 82.57 88.57 82.87 84.34 89.66 84.13
oriented interfaces. This same idea can be extended even
further to advertising. While results of integrating * P- Precision, A- Accuracy, R- Recall
personality to marketing have been mixed, some work has
demonstrated connections between marketing techniques
and consumer personality. For e-commerce marketers, both
those who advertise on Twitter and elsewhere, utilizing
social media profiles as a way to determine consumer
personality can make it easy to implement existing

© 2021, IJCSE All Rights Reserved 12

 International Journal of Computer Sciences and Engineering
Figure 4. Comparative graphs for Naïve bayes and proposed
model
V. CONCLUSION AND FUTURE SCOPE
In this paper, we have shown that users‟ Big Five
personality traits can be predicted from the public
information they share on Twitter. After processing this
data, we found many small correlations in the data. Using
the profile data as a feature set, we were able to train two
machine learning algorithms - m5sup ′Rules and Gaussian
Processes - to predict each of the five personality traits to
within 11% of its actual value. With the ability to guess a
user‟s personality traits, many opportunities are opened for
personalizing interfaces and information. We discussed
some of these opportunities for marketing and interface
design above. However, there is much work to be pursued
in this area. One area that deserves attention is the
connection between personality and the actual social
network. We considered two structural features - number
of friends and network density - but we did not look at
personality scores between friends. Understanding the
connections between personality, tie strength, trust, and
other related factors is an open space for research. By
improving our knowledge of these relationships, we can
begin to answer more sophisticated questions about how to
present trusted, socially-relevant, and well-presented
information to users.
REFERENCES
[1] Cantandir, I. Fernandez-Tobiaz, A. Belllogin, "Relating
personality types with user preferences in multiple
entertainment domains", EMPIRE 1st Workshop on Emotions
and Personality in Personalized Services, 2013.
[2] L.R. Goldberg, J.A. Johnson, H.W. Eber, R. Hogan, M.C.
Ashton, C.R. Cloninger, "The International personality item
pool and the future of public domain personality measures",
Journal of Research in Personality, vol. 40, no. 1, pp. 84-96,
2006.
[3] F. Mairesse, M. Walker, M. Mehl, R. Moore, "Using linguistic
cues for the automatic recognition of personality in
conversation and text", Journal of Artificial Intelligence
Research (JAIR), vol. 30, no. 1, pp. 457-500, 2007.
[4] Sucharitha Y, Vijayalata Y, Prasad VK. Analysis of Early
Detection of Emerging Patterns from Social Media Networks:
© 2021, IJCSE All Rights Reserved
Vol.9(4), Apr 2021, E-ISSN: 2347-2693
A Data Mining Techniques Perspective. InSoft Computing and
Signal Processing 2019 (pp. 15-25). Springer, Singapore.
[5] P.T. Costa, R.R. McCrae, "Revised NEO personality inventory
(NEO-PI-R) and NEO five-factor inventory (NEO-FFI)",
Psychological Assessment Resources, 1992
[6] Sucharitha Y, Prasad VK, Vijayalatha Y. Emergent Events
Identification in Micro-Blogging Networks Using Location
Sensitivity.
[7] J. Golbeck, C. Robles, K. Turner, "Predicting personality with
social media", CHI'11 Extended Abstracts on Human Factors
in Computing Systems, pp. 253-262, 2011. Access at ACM
[8] Yadala Sucharitha, Y. Vijayalata and V. Kamakshi Prasad*,
“Predicting Election Results from Twitter Using Machine
Learning Algorithms”, Recent Advances in Computer Science
and Communications (2020) 13: 7.
https://doi.org/10.2174/2666255813999200729164142
[9] M. Hall, E. Frank, H. Holmes, P. Pfahringer, P. Reutemann,
I.H. Witten, "The WEKA data mining software: an update",
SIGKDD Explorations, vol. 11, no. 1, 2009. Access at ACM
[10] Reddy PC, Babu AS. Survey on weather prediction using big
data analystics. In2017 Second International Conference on
Electrical, Computer and Communication Technologies
(ICECCT) 2017 Feb 22 (pp. 1-6). IEEE.
[11] N.M.A. Lestari, I.K.G.D. Putra, A.A.K.A. Cahyawan,
"Personality types classification for Indonesian text in partners
searching website using Naive Bayes methods", International
Journal of Software and Informatics Issues, 2013.
[12] Arifin, K.E. Purnaha, "Classification of emotions in Indonesia
text using K-NN method", International Journal of Information
and Electronics Engineering, vol. 2, no. 6, 2012
[13] Reddy PC, Sureshbabu A. An applied time series forecasting
model for yield prediction of agricultural crop. InInternational
Conference on Soft Computing and Signal Processing 2019 Jun
21 (pp. 177-187). Springer, Singapore.
[14] A. Firoj, E.A. Speanov, G. Riccardi, "Personality traits
recognition on social network – Facebook", ICWSM Workshop
on Personality Classification, 2013
[15] F. Iacobelli, A. Cullota, "Too neurotic not too friendly:
structured personality classification on textual data", ICWSM
Workshop on Personality Classification, 2013.
[16] Reddy PC, Babu AS. A novel approach to analysis district
level long scale seasonal forecasting of monsoon rainfall in
Andhra Pradesh and Telangana. International Journal of
Advanced Research in Computer Science. 2017 Nov 1;8(9).
[17] F. Celli, F. Pianesi, D. Stillwell, M. Kosinski, "Workshop on
computational personality recognition (shared task)",
Proceedings of WCPR13 in conjunction with ICWSM-13,
2013.
[18] Daniele Quercia§ Michal Kosinski† David Stillwell∗ Jon
Crowcroft§, Our Twitter Profiles, Our Selves: Predicting
Personality with Twitter, Horizon Digital Economy Research
grant (EP/G065802/1).
[19] Reddy PC, Sureshbabu A. An Adaptive Model for Forecasting
Seasonal Rainfall Using Predictive Analytics. InInternational
Journal of Intelligent Engineering and Systems 2019 (pp. 22-
32).
[20] Settanni M, Azucar D, Marengo D. Predicting individual
characteristics from digital traces on social media: A meta-
analysis. Cyberpsychology, Behavior, and Social Networking.
2018 Apr 1;21(4):217-28.
[21] Ozer DJ, Benet-Martinez V. Personality and the prediction of
consequential outcomes. Annu. Rev. Psychol.. 2006 Jan
10;57:401-21.
[22] Roberts BW, Kuncel NR, Shiner R, Caspi A, Goldberg LR.
The power of personality: The comparative validity of
personality traits, socioeconomic status, and cognitive ability
for predicting important life outcomes. Perspectives on
Psychological science. 2007 Dec;2(4):313-45.
[23] Soto CJ. How replicable are links between personality traits
and consequential life outcomes? The Life Outcomes of
13
 International Journal of Computer Sciences and Engineering Vol.9(4), Apr 2021, E-ISSN: 2347-2693

Personality Replication Project. Psychological Science. 2019

May;30(5):711-27.
[25] Saeb S, Lattie EG, Schueller SM, Kording KP, Mohr DC. The
relationship between mobile phone location sensor data and
depressive symptom severity. PeerJ. 2016 Sep 29;4:e2537.
[26] P Chandrashaker Reddy* and A Suresh Babu, “An Enhanced
Multiple Linear Regression Model for Seasonal Rainfall
Prediction”, International Journal of Sensors, Wireless
Communications and Control (2020) 10: 3.
[27] Ferwerda and M. Schedl, "Enhancing music recommender
systems with personality information and emotional states: A
proposal", CEUR Workshop Proc., vol. 1181, pp. 36-44, 2014.
[28] Ferwerda, E. Yang, M. Schedl and M. Tkalčič, "Personality
Traits Predict Music Taxonomy Preferences", Proc. 33rd Annu.
ACM Conf. Ext. Abstr. Hum. Factors Comput. Syst. - CHI EA
„15, pp. 2241-2246, 2015.

© 2021, IJCSE All Rights Reserved 14